0

I'm moving a large VHD file to a new VM host over the network. I've checked and double checked every point of bottlenecking here and it is 100% the server doing the writing causing an issue.

It's a RAID-10, 4x1TB Western Digital Enterprise drives on a Dell PERC6/i. It's strip size it set to 64, and and write back is enabled. I just checked the array and individual drive status in the controller BIOS and everything is healthy.

When I start the transfer speeds are nearly saturating the gigabit connection, before the cache fills and has to be written directly to disk (I think that's what is happening). Then speeds are about 10MB/s and it's looking like a ~8 hour transfer for 300GB.

I've updated the 6/i firmware, Windows is completely updated, (Server 2012R2).

EDIT: Also the new host machine (receiving the transfer) becomes so sluggish it is almost un-usable. Once the transfer is cancelled it goes back to normal.

David Makogon
  • 2,768
  • 1
  • 20
  • 29
xGlorify
  • 11
  • 3

1 Answers1

0

This can be related to cache-trashing. Try to set the disk's controller cache in write-through mode. Does this change anything?

shodanshok
  • 47,711
  • 7
  • 111
  • 180